Training samples and background points

[Alfred Chitiki]

Dear all,

On running Maxent, I am getting error messages. One indicates that some species are skipped, e.g. “Skipping Funtumia africana because it has 0 training samples” This was repeated with other species.

The other one states “No background points with data in all layers” Can anyone assist me solving these?

I will greatly appreciate

Cheers!

Alfred

[Maryam Bordkhani]

In the name of GOD

Dear Alfred

I'd suggest you  for the exactly read tutorial again and appropriate  your input layer exactly according that, then runing model again. Likely will fix your problem.


Cheers
-------------------------------------------------

Maryam Bordkhani

M.Sc.  Student of Environmental Sciences

Department of Environment, Faculty of Natural Resources,

Isfahan University of Technology, Isfahan, Iran

Email: m.bor...@gmail.com

            m.bor...@na.iut.ac.ir

[Alfred Chitiki]

Hello Maryam,
I am talking about training samples and background points and not environmental layers. In my opinion, this is a platform where Maxent users can post their problems or suggestions  BUT NOT seeking advice on how to read tutorials.

Good day

[Maryam Bordkhani]

In the name of GOD

Dear Alfred chitiki

Apparently my response has offended you!
But some time i and my friends Received many errors for due to do not steps of tutorial completely and after reading better it, errors fixed. In addition, the tutorial is NOT JUST talking about environmental layers and in this we can be found many other subjects such as background points, training samples, presence point, interpretation output and etc. For example errors about background may be related to Incorrect making bias_file.
In any case, I did not want to bother you!

your truly

---------------------------------------------

Maryam Bordkhani

M.Sc.  Student of Environmental Sciences

Department of Environment, Faculty of Natural Resources,

Isfahan University of Technology, Isfahan, Iran

Email: m.bor...@gmail.com

            m.bor...@na.iut.ac.ir

On Monday, August 5, 2013 4:44:02 AM UTC-7, Alfred Chitiki wrote:

[Alfred Chitiki]

Hello Maryam,
The simplest way was to explain whether you encountered a similar problem and how you solved it. Otherwise you could just leave the post for others to give their suggestions.

On Monday, August 5, 2013 2:44:02 PM UTC+3, Alfred Chitiki wrote:

[John Clark]

Hi Alfred,

While the problem is occurring when generating training and background points, it is nevertheless likely due to issues with your environmental variables.  If your previous description, "Maxent managed to run the model successfully when using the layers derived from a similar source", is still relevant, then you should take a closer look at your prepared variables.  In ArcGIS, even with proper mask and snap raster environmental settings, the pixels with the same resolution may still not stack properly.  Zoom all the way in and make sure the pixels are PERFECTLY aligned.  I found that I could only get ArcGIS to consistently honor the snap raster by using ArcPy.  You can also check the header files to make sure the extents are consistent.  If not, try to re-clip using the smallest.  You may also have variables with a few scattered "NoData" holes within the study area.  Finally, assure your presence points are in the right projection.  Circumvent potential confusions with on-the-fly projection by comparing the actual x,y coordinates in your .csv file to the extent coordinates in your raster header.  Hope that helps.  It looks like Maryam was pointing you in the right direction by suggesting you reconsider your variable pre-processing steps, but language barriers may have obscured their intent.  If you're still having trouble I'd suggest searching the group's archives as this is among the most frequent questions asked (along with variable selection and model evaluation.)  Good luck!

Best Regards,

John Clark

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To unsubscribe from this group and stop receiving emails from it, send an email to maxent+un...@googlegroups.com.
To post to this group, send email to max...@googlegroups.com.
Visit this group at http://groups.google.com/group/maxent.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[rprieto]

Hi Alfred,

I believe John is correct. Aligning the rasters in Arc can be a big headache. I've ran in many problems when trying to do that with rasters from different origins and with diverse projections and resolutions. Sometimes the rasters would not align by a few meters.

When that is the case, one workaround to which I resourced often is to transform the rasters to points and then back to rasters. This is easy to do with small to medium sized rasters, but will be time consuming if your rasters are large (in other words, if you are working in a large area).

I am attaching the workflow for the model I created to do that in the Arc Model Builder. The process is simple, but assumes that the resolution of the rasters is the same:
1- Using the raster that you want to serve as reference for the alignment, create a grid of points (using conversion tools);
2- Open the attribute table of the resulting point layer and remove the columns with values. This will clean your reference point grid;
3- Use the extract values to points tool (if you have many rasters you can do that through a process as the one in the image or by using the tool in batch);
4- Convert your grids of points back to rasters (using the conversion tools, in batch if necessary).

Since the point grid used to sample the rasters is always the same, the resulting rasters will align perfectly. Since the point grid was created from a reference raster (and assuming your rasters are not aligning by only a few meters) the sampling will be done near the centre of the rasters, so there is no chance that you will sample the wrong cell.

There are other ways of aligning rasters but this process works every time. Nevertheless, I have to stress again that it may be very slow if your rasters are large. In that case, consider other options, such as using ArcPy to make Arc honour the snap raster, as suggested by John.

good luck,

Rui Prieto

[Alfred Chitiki]

Dear John and Rui,
Many thanks for your suggestions. I will take your advice in action. I will get back for the feedback.

Cheers guys!

Alfred

On Monday, August 5, 2013 2:44:02 PM UTC+3, Alfred Chitiki wrote:

[Alfred Chitiki]

Hi Guys,

Following the error messages on training and background samples, I checked all ascii layers on a notepad. After noting the differences between header lines of environmental variables, Using Arc map, I modified the layers using Extract by Mask tool in Spatial Analyst. For the Input raster, I used a variable of interest (i.e. a variable that need to be modified). For the Input raster or feature mask data field, I used a standard raster layer (i.e. the extent you wish to match). I have set the coordinate system, cell size and extent, on the Environments… button. For the variables Output Coordinates, again I used a variable of interest. For the Processing Extent and Snap Raster variables, again I used a standard raster layer. For the field of Cell Size under the Raster Analysis variable, I entered the environmental layer that has the smallest cell.

It worked quite well.

Ciao!

Alfred

On Monday, August 5, 2013 2:44:02 PM UTC+3, Alfred Chitiki wrote: